gianluca.verin verin@libero. @libero.itit Vicenza.linux.it\LinuxCafe 1
Agenda IPv6 Basics Connecting to 6Bone Why do we need IPv6? IPv6 Introduction-Transition IPv6 and open source community Future applications Summary Vicenza.linux.it\LinuxCafe 2
Introduction to IPv6 IP Standards Body - IETF (Internet Engineering Task Force) IETF recognised the need to work on an improved Internet Protocol IP Next Generation (IPng) Directorate formed in 1991 More than 10 years of work on IPng resulted in IPv6 IPv6 is now mature all new IP protocols developed in IETF are now expected to support IPv6 IETF/IAB has recommended the use of IPv6 for wireless (billions of devices) IPv6 adopted by 3GPP for IP Multimedia services Government Actions Japan (incentives for IPv6) and China (NGN network) U.S. DoD (Network-centric warfare) EU IPv6 Task Force (IPv6 is part of the e-europe Broadband Action) Vicenza.linux.it\LinuxCafe 3
The IPv4 Header (>= 20 bytes) 0 bits 4 8 16 24 31 Ver HL Type of Service Total Length Identifier Flags Fragment Offset Time to Live Protocol Header Checksum 32 bit Source Address 32 bit Destination Address Options and Padding Variable header length (e.g. for options) and fragmenting makes it harder to do fast processing in software Checksum redundant since error checks always done at layer-2 Vicenza.linux.it\LinuxCafe 4
The IPv6 Header (40 bytes) 0 4 12 16 24 31 Version Traffic Class Flow Label Payload Length Next Header Hop Limit 128-bit Source Address 128-bit Destination Address Version 6 Traffic Class Priority of IPv6 packets Flow Label Special handling of packet Payload Length Next Header Hop Limit Length of IPv6 payload Type of header following the IPv6 header Decremented by 1 in each router Vicenza.linux.it\LinuxCafe 5
Extension Headers IPv6 Header Next Header = TCP TCP Header + Data IPv6 Header Next Header = Hop-by-Hop Hop-by-Hop Header Next Header = UDP UDP Header + Data IPv6 Header Next Header = Destination Destination Header Next Header = Routing Routing Header Next Header = ESP ESP Header Next Header = TCP TCP Header + Data Vicenza.linux.it\LinuxCafe 6
IPv6 Addresses There are three different types of addresses: Unicast - Anycast - An identifier for a single interface (e.g. Aggregatable Global Unicast) An identifier for a set of interfaces. A packet sent to an Anycast address is delivered to one of the interfaces identified by that address. Anycast addresses could be used to reach the nearest node for a certain service. Multicast - An identifier for a set of interfaces. A packet sent to a multicast address is delivered to all interfaces identified by that address. Vicenza.linux.it\LinuxCafe 7
IPv6 Unicast Address scopes Link-local Only unique/valid on same link (ALL DEVICES NEED THIS) Autconfigured at startup. Packets are not forwarded beyond the link when the source or destination is link-local Site-local (DEPRECATED) Only unique/valid within the same site (operator-defined) May be autoconfigured. Packets are not forwarded beyond the site when the source or destination is site-local Global Globally unique. May be autoconfigured. Vicenza.linux.it\LinuxCafe 8
IPv6 Aggregatable Global Unicast Addressing Service Provider Service Provider Service Provider Subnet ID (n bits) Global Prefix Interface ID Site Site Site Link Locator (routing part) 8 bytes Interface Identifier 8 bytes Geographical significance of IPv6 unicast addressing allows efficient route aggregation Example IPv6 assignment: /32 => 32-bit subnet id Stateless example: Each /32 can be broken up into > 4.2 Billion /64s distributed to links/hosts Vicenza.linux.it\LinuxCafe 9
IPv6 Addresses :: unspecified address ::1 loopback address fe80::<64bits interface id> Link local address 2001:923f:200:101::/64. Globally routable IPv6 prefix assigned to hosts connected to a router s interface 2001:923f:200::/48. Typical prefix assigned to networks. Space for 2^16 = 65K /64 prefix Special format for globally routable address 2002:v4_addr::/48 6-to-4 prefix </64 prefix>:0000:5efe:v4_addr ISATAP address Vicenza.linux.it\LinuxCafe 10
Applying for IPv6 Addresses Operator should apply for a /32 IPv6 Global prefix to its local Regional or National Internet Registry (RIR/NIR): RIPE for Europe RIPE (www.ripe.net) ftp://ftp.ripe.net/ripe/docs/ripe-267.txt Else follow step by step instruction in SixXS in www.sixxs.net/main/. SixXS offers tunnel broker service and upon request you usually get a /48 prefix. Vicenza.linux.it\LinuxCafe 11
Connecting to 6Bone If your ISP does not offer native IPv6 connectivity you may register with 6Bone and use a tunnel broker such as SixXS. From the tunnel broker you get: IPv4 address of tunnel broker end point, IPv6 address ISP end point, your IPv6 address. Optionally a /48 routable prefix FreeBSD configuration example: # ifconfig gif0 create # ifconfig gif0 tunnel [Your IPv4 Endpoint] [POP IPv4 Endpoint] # ifconfig gif0 inet6 [Your IPv6 Endpoint] [POP IPv6 Endpoint] prefixlen 128 Route your IPv6 traffic via the SixXS POP: # route add -inet6 default [POP IPv6 Endpoint] If you plan to use your machine as a router: # sysctl -w net.inet6.ip6.forwarding=1 Send RA to interface (prefix contained in file /etc/rtadvd.conf ) #rtadv [interface name] Vicenza.linux.it\LinuxCafe 12
What do IPv6 addresses look like? Tunnel Interface ifconfig gif0 IPv4 pop address gif0: flags=8051<up,pointopoint,running,multicast> mtu 1280 Global tunnel inet 192.168.1.3 --> 212.224.0.188 Unicast inet6 2001:6f8:900:43f::2 --> 2001:6f8:900:43f::1 prefixlen 128 inet6 fe80::208:74ff:fee9:baa8%gif0 prefixlen 64 scopeid 0x5 Link-Local Unicast Vicenza.linux.it\LinuxCafe 13
IPv6 Address Configuration Stateless (server-less) Does not depend on communication with a server (reliability increased) IPv6 hosts automatically configure addresses without the need for user intervention (Concept is described in RFC2462) For those services where user authentication & accounting is needed, RADIUS server may be used also to allocate prefixes Stateful Automatic address configuration given to hosts by DHCPv6 server Requries a DHCPv6 server in the network DHCPv6 client in host, first-hop router as DHCPv6 relay Vicenza.linux.it\LinuxCafe 14
IPv6 Neighbor Discovery (Stateless Addressing) 1. Host forms Link-local address (combining link-local prefix and interface Id) PPP, Ethernet 2. Host optionally sends Router Solicitation (link-local multicast) IPv6 Host 3. Router sends periodic multicast Router Advertisement or unicast in response to solicitation 4. Host forms Global address (128 bits) by combining First-hop (default) IPv6 router Global Prefix Interface ID Neighbor Discovery (ND) uses ICMPv6 Router Advertisement (RA) contains Global Prefix, Lifetimes Interface Identified (IID) may be based on MAC address or random number Vicenza.linux.it\LinuxCafe 15
IPv6 Neighbor Discovery (continued) 1. Neighbor Solicitation 2. Neighbor Advertisement First-hop (default) IPv6 router IPv6 Host1 1. Neighbor Solicitation (who has address xyz?) (solicited node multicast) IPv6 Host2 (on-link) 2. Neighbor Advertisement (unicast) ND replaces ARP (IPv4) for link-layer address resolution of nodes on the same IPv6 link link-local addresses are used for these messages Neighbor Unreachability Detection (NUD) to determine when a neighbor is no longer reachable (e.g. default router) Vicenza.linux.it\LinuxCafe 16
How are IPv6 addresses assigned to Mobiles? Mobile Terminal /64 Global Prefix Terminal Equipment (e.g. Laptop) PDP Context Activation IPv6 Link-local Router Advertisement GGSN (Default Router) GPRS/UMTS Operator network Pool of /64 prefixes per APN Local or remote (RADIUS) IPv6 Internet IETF/3GPP successful collaboration produced future-proof IPv6 standard for Mobile Networks Mobile host s Link-local address provided by GGSN to avoid duplication Each Mobile Terminal is assigned a unique /64 IPv6 prefix which can be used to create multiple addresses (privacy), Personal Area Networks etc. Unreliable DAD may be avoided (reduces messages over air) Vicenza.linux.it\LinuxCafe 17
A few IPv6 Specifications IPv6 Protocol Specification (RFC2460) IPv6 Neighbor Discovery (RFC2461) being updated IPv6 Stateless Address Autoconfiguration (RFC2462) being updated IPv6 Internet Control Message Protocol (RFC2463) being updated IPv6 Addressing Architecture (RFC3513) IPv6 over Ethernet (RFC2464) Support for IPv6 in Session Description Protocol (SDP) (RFC 3266) Mobile IPv6 (RFC 3775) Dynamic Host Configuration Protocol for IPv6 (RFC 3315) Recommendations for IPv6 in 3GPP Standards (RFC 3314) Internet Protocol Version 6 (IPv6) for Some Second and Third Generation Cellular Hosts (RFC 3316) Transition Scenarios for 3GPP Networks (RFC 3574) Vicenza.linux.it\LinuxCafe 18
Market Trends & Demands World Mobile Subscriptions forecast to pass the 2 Billion mark in 2007 * Restrictive IPv4 Assignment Policies IPv4 Unassigned addresses expected to run out in 2018 Not enough IPv4 public addresses for mobile users Need a long-term solution for continued market growth Traditionally when more circuit switched phone numbers are needed this is solved by modifying/adding area codes We don t share phone numbers, why share IP addresses? Increased network security: working e2e security model Increased Quality of Service demands * Ovum Sept 2004 Vicenza.linux.it\LinuxCafe 19
Why can t we just use IPv4 NATs? NATs translate IP addresses (including IP addresses used inside applications when the NAT is combined with an ALG) Security is problematic with NATs NATs need to process all packets => can add delay to delay-sensitive sensitive packets (e.g. voice, video) NATs keep per-connection state => scalability is an issue What if a NAT fails? => Reliability Need to support ALL applications which use IP addresses at the application level (FTP, Netmeeting etc.) => What happens to new or proprietary/secret applications? NATs do not provide security on their own need a Firewall IPv6 allows you to do away with these problems Vicenza.linux.it\LinuxCafe 20
IPv6 Other advantages Built-in security support (IPsec) Efficient routing Features for simplified Network operations Address autoconfiguration Automated network (router) renumbering Automated Server discovery (e.g. DNS) Same level of Quality of Service (QoS) support as in IPv4 with potential improvement using Flow Label Built in mobility support (MIPv6) Vicenza.linux.it\LinuxCafe 21
End-to-end incompatibility IPv6 host communicating with an IPv4 peer V6 host End-to-end V4 host IPv6 network IP-backbone IPv4 network Solutions: Dual stack networks (but then every network on the left should have both an IPv6 and IPv4 address, not always possible if the host is a mobile) Translators: Stateful Translators (NA(P)T-PT), with ALGs Stateless Translators (SIIT) not recommended Dual-stack Application Proxies (e.g. HTTP, FTP, E-mail, WAP) Vicenza.linux.it\LinuxCafe 22
Automatic tunnelling, using heartbit V6 Service IPv4 Network v6 in v4 IPv6 domain V6/V4 host Heartbeat Dual stack router V6/V4 host change IPv4 address, need to communicate host IPv4 address. Procedure password protected Heartbeat every 60 seconds, after 300 seconds tunnel tear down Vicenza.linux.it\LinuxCafe 23
IPv6 Support in OS Virtually every OS that runs on PC and server supports IPv6. Linux FreeBSD, NetBSD, OpenBSD But also Solaris HP-UX Windows Vicenza.linux.it\LinuxCafe 24
Open Source IPv6 Applications Many of today open source applications are dual stacks. Just to list a few: Apache www server Firefox www client Ircd irc server Xchat2 irc client ftp, telnet, openssh. Vicenza.linux.it\LinuxCafe 25
Open Source supporting IPv6 Kame www.kame.net. IPv6 open source stack ported to BSD main tree USAGI www.linux-ipv6.org. Patch to Linux IPv6 kernel. TAHI http://www.tahi.org/. Set of test to assure conformance and interoperability. Collaborate in the IPv6 ready logo program Nautilus6 http://www.nautilus6.org/ IPv6 mobility related technologies Ferrara LUG has a web page on IPv6 Vicenza.linux.it\LinuxCafe 26
IPv6 socket API Application developers are required to support both IPv4 and IPv6 Basic Socket Interface Extensions for IPv6 (RFC 3493) Advanced Sockets Application Program Interface (API) for IPv6 (RFC 3542) Porting of application from IPv4 to IPv6 require changes in Structures used (e.g. sockaddr_in6 instead of sockaddr_in) INADDR_ANY and INADDR_LOOPBACK are not IPv6 compatible and need to be replaced Protocol family AF_INET and PF_INET are replaced by AF_INET6 and PF_INET6 Some function like gethostbyname() that only support IPv4 should be replaced by functions such as getaddrinfo() that supports both IPv4 and IPv6 Vicenza.linux.it\LinuxCafe 27
IPv6 future usage: Prefix Assignment Mobile Terminal Personal Area Network /64 Global Prefix Assignment IETF/3GPP successful collaboration produced future-proof IPv6 standard for Mobile Networks Each Mobile Terminal is assigned a unique /64 IPv6 prefix Personal Area Networks and Moving Networks Privacy addressing Vicenza.linux.it\LinuxCafe 28
Future usage of IPv6 in Moving Networks Connection Activation IPv6 Link-local Mobile Network GGSN/PDSN IPv6 Services Network Router Advertisement Router Advertisement /64 Global Prefix IPv6 Internet Vicenza.linux.it\LinuxCafe 29
Summary IPv6 is not backward compatible with IPv4 The area interested by the transition is anything has to with the internet today and tomorrow from applications to networking. The community for open source has been very active in developing applications and stacks for IPv6 Now is the time to get it spread and start to use the new Internet and its new possibilities IPv6 will enable the direct communication between hosts in the Internet and reachability Vicenza.linux.it\LinuxCafe 30